covid-19

A GraphQL server on top of a MongoDB database for COVID-19 time series data for counties in the United States.

Table of Contents

• Querying the Production Server
  • Additional Methods & Examples for Querying the Heroku Server
    • GET request
    • POST request
• Local Setup & Development
  • Requirements
  • Clone and Install
  • Environment Variables
  • Seeding the Database
  • Starting the GraphQL Server
• The Data
  • Fetching & Parsing
  • Transforming & Cleaning

---

Querying the Production Server

A production version of the GraphQL server is hosted on Heroku. The endpoint is at: https://covid-19-73586.herokuapp.com/ no longer active. A scheduler add-on runs the seed script once a day to update the MongoDB Atlas database cluster.

If you have Postman feel free to import the collection and play around with the different ways to send requests.

Additional Methods & Examples for Querying the Heroku Server

Each example below sends the following GraphQL query and variables.

query($where: EntriesWhereInput!) {
  entries(where: $where) {
    id
    combined_name
    population
    loc {
      type
      coordinates
    }
    date
    confirmed
  }
}

{
  "where": {
    "state": "Kentucky",
    "county": "Jefferson"
  }
}

GET request

GET /?query=query(%24where%3A%20EntriesWhereInput!)%20%7B%20entries(where%3A%20%24where)%20%7B%20id%20combined_name%20population%20loc%20%7B%20type%20coordinates%20%7D%20date%20confirmed%20%7D%20%7D&variables=%7B%22where%22%3A%7B%22state%22%3A%22Kentucky%22%2C%22county%22%3A%22Jefferson%22%7D%7D HTTP/1.1
Host: covid-19-73586.herokuapp.com
Content-Type: application/json

cURL

curl 'https://covid-19-73586.herokuapp.com/?query=query(%24where%3A%20EntriesWhereInput!)%20%7B%20entries(where%3A%20%24where)%20%7B%20id%20combined_name%20population%20loc%20%7B%20type%20coordinates%20%7D%20date%20confirmed%20%7D%20%7D&variables=%7B%22where%22%3A%7B%22state%22%3A%22Kentucky%22%2C%22county%22%3A%22Jefferson%22%7D%7D' \
  --request GET \
  --header 'Content-Type: application/json'

JavaScript - Fetch

const getHeaders = new Headers({ 'Content-Type': 'application/json' })

const requestOptions = {
  method: 'GET',
  headers: getHeaders,
  redirect: 'follow',
}

fetch(
  'https://covid-19-73586.herokuapp.com/?query=query(%24where%3A%20EntriesWhereInput!)%20%7B%20entries(where%3A%20%24where)%20%7B%20id%20combined_name%20population%20loc%20%7B%20type%20coordinates%20%7D%20date%20confirmed%20%7D%20%7D&variables=%7B%22where%22%3A%7B%22state%22%3A%22Kentucky%22%2C%22county%22%3A%22Jefferson%22%7D%7D',
  requestOptions
)
  .then((response) => response.text())
  .then((result) => console.log(result))
  .catch((error) => console.log('error', error))

POST request

POST / HTTP/1.1
Host: covid-19-73586.herokuapp.com
Content-Type: application/json
Content-Length: 208

{"query":"query($where: EntriesWhereInput!) { entries(where: $where) { id combined_name population loc { type coordinates } date confirmed } }","variables":{"where":{"state":"Kentucky","county":"Jefferson"}}}

cURL

curl 'https://covid-19-73586.herokuapp.com/' \
  --request POST \
  --header 'Content-Type: application/json' \
  --data-raw '{"query":"query($where: EntriesWhereInput!) { entries(where: $where) { id combined_name population loc { type coordinates } date confirmed } }","variables":{"where":{"state":"Kentucky","county":"Jefferson"}}}'

JavaScript - Fetch

const postHeaders = new Headers({ 'Content-Type': 'application/json' })

const graphql = JSON.stringify({
  query: `query($where: EntriesWhereInput!) {
    entries(where: $where) {
      id
      combined_name
      population
      loc {
        type
        coordinates
      }
      date
      confirmed
    }
  }`,
  variables: {
    where: {
      state: 'Kentucky',
      county: 'Jefferson',
    },
  },
})

var requestOptions = {
  method: 'POST',
  headers: postHeaders,
  body: graphql,
  redirect: 'follow',
}

fetch('https://covid-19-73586.herokuapp.com', requestOptions)
  .then((response) => response.text())
  .then((result) => console.log(result))
  .catch((error) => console.log('error', error))

Local Setup & Development

Requirements

• Node.js v13.9.0 or later. But should work with v12.18.2 or later (haven't tested it yet though).

• To run and test the database locally you will need MongoDB. I've been using v4.2.8.

Clone and Install

Clone the repo and install the dependencies through npm.

git clone https://github.com/klazich/covid-19
cd covid-19
npm install

Environment Variables

The graphql server and mongo database require certain environment variables to be set before startup. If they are not defined than the app will fail to start. Make a copy of the .env-example (located at project root, here) and rename it to .env and place it in the root of the project.

PORT=4000
MONGODB_URL=mongodb://localhost:27017/covid-19
NODE_ENV=development

With dotenv-safe the correct development environment variables will be set when starting up the server and database.

Seeding the Database

If you use a free database service, such as mLab's free tier, inserting the full dataset WILL consume a large portion of your quota. See below on how to limit the number of documents created.

Seed the database with data by running:

npm run seed

This will initiate the fetching, parsing, transforming and inserting of the data into the mongo database.

You can pass a limit option to the script to limit how many documents are created and saved to the database. For example, to limit the number of documents created to the first ten thousand you would enter:

npm run seed -- --limit=10000

This can be helpful when debugging database connection issues as seeding the database with the full dataset will create over 500,000 documents and can take up to 30 seconds.

Starting the GraphQL Server

To start the GraphQl server run:

npm start

This will start up the development server locally (when NODE_ENV is 'development') at: http://localhost:4000/, and connect to the mongodb database.

Be sure to seed the database (see above) at some point. Otherwise queries will return empty.

The development server also spins up a graphql playground that can be accessed by navigating your browser to the endpoint (http://localhost:4000/). The playground will load with some example queries that you can go ahead and execute.

The Data

The COVID-19 data is sourced from the COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University. More specifically, the confirmed US cases time series csv table and the FIPS lookup table.

See the database directory, here for the relaxant code.

Fetching & Parsing

The data starts in a csv file with headers looking like this:

UID	iso2	iso3	code3	FIPS	Admin2	Province_State	Country_Region	Lat	Long_	Combined_Key	...	7/16/20	7/17/20	7/18/20	7/19/20	7/20/20	7/21/20	7/22/20	7/23/20	7/24/20	7/25/20	...
84021111	US	USA	840	21111.0	Jefferson	Kentucky	US	38.18664655	-85.65931031	"Jefferson, Kentucky, US"	...	4778	4861	4962	5170	5263	5441	5519	5635	5836	6080	...

Using axios to create an http data stream and papaparse to parse the csv data, each row is transformed into a javascript object:

{
  UID: '84021111',
  iso2: 'US',
  iso3: 'USA',
  code3: '840',
  FIPS: '21111.0',
  Admin2: 'Jefferson',
  Province_State: 'Kentucky',
  Country_Region: 'US',
  Lat: '38.18664655',
  Long_: '-85.65931031',
  Combined_Key: 'Jefferson, Kentucky, US',
  // ...
  '7/16/20': '4778',
  '7/17/20': '4861',
  '7/18/20': '4962',
  '7/19/20': '5170',
  '7/20/20': '5263',
  '7/21/20': '5441',
  '7/22/20': '5519',
  '7/23/20': '5635',
  '7/24/20': '5836',
  '7/25/20': '6080',
  // ...
}

Transforming & Cleaning

From the above javascript object, we create a new object for each of the date keys along with a copy of all the county's information. These objects will eventually be the documents inserted into the MongoDB database collection. The keys are also cleaned and the values correctly typed. In the end the object conforms to the mongoose schema and the graphql schema. Continuing from above, here is an example:

{
  uid: 84021111,
  country_iso2: 'US',
  country_iso3: 'USA',
  country_code: 840,
  fips: 21111,
  county: 'Jefferson',
  state: 'Kentucky',
  country: 'US',
  combined_name: 'Jefferson, Kentucky, US',
  loc: {
    type: 'Point',
    coordinates: [ -85.65931031, 38.18664655 ]
  },
  population: 766757,
  date: 2020-07-25T04:00:00.000Z,
  confirmed: 6080,
}